Pro 200 resets and “warm cache” effect

A user is testing Codex on Pro 200 and has two main questions. First, they want to know if there's a "warm cache" effect where quota usage becomes cheaper in longer sessions as more context is cached. They are asking if anyone has measured this. Second, they are inquiring about banked usage resets for Pro 200, specifically if a full reset restores the allowance applicable at that moment, especially with the upcoming reduction from 20x to 10x.